Order Book Prediction Model

Deep learning model for predicting values from order book data using Transformer and LSTM architectures.

Features

• Order book feature engineering
• Transformer and LSTM model implementations
• Time series k-fold cross-validation
• TensorBoard integration
• GPU support

Project Structure

├── src/
│   ├── dataloader.py         # Data preprocessing
│   ├── LSTM_utils.py         # LSTM model
│   ├── transformer_utils.py  # Transformer model
│   └── utils.py             # Utility functions
├── inference_model.py        # Model inference
├── k-fold.py                # Cross validation
└── requirements.txt         # Dependencies

Usage

Install dependencies:

pip install -r requirements.txt

Run inference:

python inference_model.py <path_to_csv>

Configuration

Model parameters can be adjusted in the Config class:

• Window size and batch size
• Model architecture (Transformer/LSTM)
• Training parameters (learning rate, epochs)
• Feature reduction options